Datasets:

sswt

/

arxiv_sample_50K

like 0

Minimal Neutral Naturalness Model ; We build a minimal neutral naturalness model in which the top partners are not charged under QCD, with a pseudo Goldstone Higgs arising from SO5SO4 breaking. The colorneutral top partners generate the Higgs potential radiatively without quadratic divergence. The misalignment between the electroweak scale and global symmetry breaking scale is naturally obtained from suppression of the Higgs quadratic term, due to cancellation between singlet and doublet top partner contributions. This model can be embedded into ultraviolet holographic setup in composite Higgs framework, which even realizes finite Higgs potential.

Learning Noun Cases Using Sequential Neural Networks ; Morphological declension, which aims to inflect nouns to indicate number, case and gender, is an important task in natural language processing NLP. This research proposal seeks to address the degree to which Recurrent Neural Networks RNNs are efficient in learning to decline noun cases. Given the challenge of data sparsity in processing morphologically rich languages and also, the flexibility of sentence structures in such languages, we believe that modeling morphological dependencies can improve the performance of neural network models. It is suggested to carry out various experiments to understand the interpretable features that may lead to a better generalization of the learned models on crosslingual tasks.

Multiclass Classification Model Inspired by Quantum Detection Theory ; Machine Learning has become very famous currently which assist in identifying the patterns from the raw data. Technological advancement has led to substantial improvement in Machine Learning which, thus helping to improve prediction. Current Machine Learning models are based on Classical Theory, which can be replaced by Quantum Theory to improve the effectiveness of the model. In the previous work, we developed binary classifier inspired by Quantum Detection Theory. In this extended abstract, our main goal is to develop multiclass classifier. We generally use the terminology multinomial classification or multiclass classification when we have a classification problem for classifying observations or instances into one of three or more classes.

Learning to Jointly Translate and Predict Dropped Pronouns with a Shared Reconstruction Mechanism ; Pronouns are frequently omitted in prodrop languages, such as Chinese, generally leading to significant challenges with respect to the production of complete translations. Recently, Wang et al. 2018 proposed a novel reconstructionbased approach to alleviating dropped pronoun DP translation problems for neural machine translation models. In this work, we improve the original model from two perspectives. First, we employ a shared reconstructor to better exploit encoder and decoder representations. Second, we jointly learn to translate and predict DPs in an endtoend manner, to avoid the errors propagated from an external DP prediction model. Experimental results show that our approach significantly improves both translation performance and DP prediction accuracy.

Neural Transitionbased Syntactic Linearization ; The task of linearization is to find a grammatical order given a set of words. Traditional models use statistical methods. Syntactic linearization systems, which generate a sentence along with its syntactic tree, have shown stateoftheart performance. Recent work shows that a multilayer LSTM language model outperforms competitive statistical syntactic linearization systems without using syntax. In this paper, we study neural syntactic linearization, building a transitionbased syntactic linearizer leveraging a feedforward neural network, observing significantly better results compared to LSTM language models on this task.

Minimal gauge inflation and the refined Swampland conjecture ; The refined de Sitter dS conjecture provides two consistency conditions for an effective theory potential of a quantum gravity theory. Any inflationary model can be checked by these conditions and minimal gauge inflation is not an exception. We develop a generic method to analyze a monotonically growing potential with an inflection point on the way to the plateau near the top such as the potential in minimal gauge inflation model and the Higgs inflation. Taking the latest observational data into account, we find the fully consistent parameter space where the model resides in the Landscape rather than in the Swampland.

On the long time behavior of a tumor growth model ; We consider the problem of the long time dynamics for a diffuse interface model for tumor growth. The model describes the growth of a tumor surrounded by host tissues in the presence of a nutrient and consists in a CahnHilliardtype equation for the tumor phase coupled with a reactiondiffusion equation for the nutrient concentration. We prove that, under physically motivated assumptions on parameters and data, the corresponding initialboundary value problem generates a dissipative dynamical system that admits the global attractor in a proper phase space.

StressTesting Neural Models of Natural Language Inference with MultiplyQuantified Sentences ; Standard evaluations of deep learning models for semantics using naturalistic corpora are limited in what they can tell us about the fidelity of the learned representations, because the corpora rarely come with good measures of semantic complexity. To overcome this limitation, we present a method for generating data sets of multiplyquantified natural language inference NLI examples in which semantic complexity can be precisely characterized, and we use this method to show that a variety of common architectures for NLI inevitably fail to encode crucial information; only a model with forced lexical alignments avoids this damaging information loss.

ZeroShot Anticipation for Instructional Activities ; How can we teach a robot to predict what will happen next for an activity it has never seen before We address this problem of zeroshot anticipation by presenting a hierarchical model that generalizes instructional knowledge from largescale textcorpora and transfers the knowledge to the visual domain. Given a portion of an instructional video, our model predicts coherent and plausible actions multiple steps into the future, all in rich natural language. To demonstrate the anticipation capabilities of our model, we introduce the Tasty Videos dataset, a collection of 2511 recipes for zeroshot learning, recognition and anticipation.

Macroscopic analysis of shotnoise Cox random balls ; In this paper, we consider a cluster model of weighted Euclidean random balls generated by a shotnoise Cox process. It is an example of cluster point process. We perform a scaling on the model by shrinking the radii of the balls and compensate this effect by increasing the mean number of balls in each cluster, orand increasing the mean number of clusters. We consider two different scenarios, say a local and a global scenarios. Heuristically, in the first scenario, we focus on the mean number of large balls in a cluster while in the second one, we focus on the global mean number of large balls in the model. According to the different scenarios, the cluster structure can persist at the limit or disappear.

Seven largest couplings of the standard model as IR fixed points ; We report on an intriguing observation that the values of all the couplings in the standard model except those related to first two generations can be understood from the IR fixed point structure of renormalization group equations in the minimal supersymmetric model extended by one complete vectorlike family with the scale of new physics in a multiTeV range.

Marginal deformations of WZW models and the classical YangBaxter equation ; We show how socalled YangBaxter YB deformations of sigma models, based on an Rmatrix solving the classical YangBaxter equation CYBE, give rise to marginal currentcurrent deformations when applied to the WessZuminoWitten WZW model. For noncompact groups these marginal deformations are more general than the ones usually considered, since they can involve a nonabelian current subalgebra. We classify such deformations of the AdS3 x S3 string.

Composite gravity from a metricindependent theory of fermions ; We present a metricindependent, diffeomorphisminvariant model with interacting fermions that contains a massless composite graviton in its spectrum. The model is motivated by the supersymmetric Dbrane action, modulated by a fermion potential. The gravitational coupling is related to new physics at the cutoff scale that regularizes UV divergences. We also speculate on possible extensions of the model.

Modelling the Incomplete Intermodal Terminal Location Problem ; In this paper, we introduce and study the incomplete version of the intermodal terminal location problem. It's a generalization of the classical version by relaxing the assumption that the induced graph by located terminals is complete. We propose a mixed integer program to model the problem and we provide several extensions. All models are tested through validation in CPLEX solver. Numerical results are reported using wellknown data set from the literature.

Does gravity have to be quantized Lessons from nonrelativistic toy models ; It is often argued that gravity has to be a quantum theory simply because a fundamentally semiclassical approach would necessarily be inconsistent. Here I review recent Newtonian toy models of stochastic semiclassical gravity. They provide one option to implement a force semiclassically without getting into the known problems associated with meanfield. These models are not complete theories and should not be considered too seriously, but their consistency shows that semiclassical gravity is hard to dismiss on purely theoretical grounds.

Master Majorana neutrino mass parametrization ; After showing that the neutrino mass matrix in all Majorana models can be described by a general master formula, we will present a master parametrization for the Yukawa matrices, also valid for all Majorana models, that automatically ensures agreement with neutrino oscillation data. The application of the master parametrization will be illustrated in an example model.

Aesthetics of Neural Network Art ; This paper proposes a way to understand neural network artworks as juxtapositions of natural image cues. It is hypothesized that images with unusual combinations of realistic visual cues are interesting, and, neural models trained to model natural images are wellsuited to creating interesting images. Art using neural models produces new images similar to those of natural images, but with weird and intriguing variations. This analysis is applied to neural art based on Generative Adversarial Networks, image stylization, Deep Dreams, and Perception Engines.

cal PT deformation of CalogeroSutherland models ; CalogeroSutherland models of N identical particles on a circle are deformed away from hermiticity but retaining a cal PT symmetry. The interaction potential gets completely regularized, which adds to the energy spectrum an infinite tower of previously nonnormalizable states. For integral values of the coupling, extra degeneracy occurs and a nonlinear conserved supersymmetry charge enlarges the ring of Liouville charges. The integrability structure is maintained. We discuss the AN1type models in general and work out details for the cases of A2 and G2.

Permutation patterns in genome rearrangement problems the reversal model ; In the context of the genome rearrangement problem, we analyze two well known models, namely the reversal and the prefix reversal models, by exploiting the connection with the notion of permutation pattern. More specifically, for any k, we provide a characterization of the set of permutations having distance leq k from the identity which is known to be a permutation class in terms of what we call generating peg permutations and we describe some properties of its basis, which allow to compute such a basis for small values of k.

Transferability of Operational Status Classification Models Among Different Wind Turbine Typesq ; A detailed understanding of wind turbine performance status classification can improve operations and maintenance in the wind energy industry. Due to different engineering properties of wind turbines, the standard supervised learning models used for classification do not generalize across data sets obtained from different wind sites. We propose two methods to deal with the transferability of the trained models first, data normalization in the form of power curve alignment, and second, a robust method based on convolutional neural networks and featurespace extension. We demonstrate the success of our methods on realworld data sets with industrial applications.

Existence and Regularity of Weak Solutions for a Thermoelectric Model ; This paper concerns a timeindependent thermoelectric model with two different boundary conditions. The model is a nonlinear coupled system of the Maxwell equations and an elliptic equation. By analyzing carefully the nonlinear structure of the equations, and with the help of the De GiorgiNash estimate for elliptic equations, we obtain existence of weak solutions on Lipschitz domains for general boundary data. Using Campanato's method, we establish regularity results of the weak solutions.

Unified Superradiant phase transitions ; We prove, by means of a unified treatment, that the superradiant phase transitions of Dicke and classical oscillator limits of simple lightmatter models are indeed of the same type. We show that the meanfield approximation is exact in both cases, and compute the structure and location of the transitions in parameter space. We extend this study to a fuller range of models, paying special attention to symmetry considerations. We uncover general features of the phase structure in the space of parameters of these models.

Growing graphs with addition of communities ; Paper proposes a model of large networks based on a random preferential attachment graph with addition of complete subgraphs cliques. The proposed model refers to models of random graphs following the nonlinear preferential attachment rule and takes into account the possibility of guillemotleftaddingguillemotright entire communities of nodes to the network. In the derivation of the relations that determine the vertex degree distribution, the technique of finitedifference equations describing stationary states of a graph is used. The obtained results are tested empirically by generating large graphs, special cases correspond to known mathematical relations.

A General Framework for the Semantics of Type Theory ; We propose an abstract notion of a type theory to unify the semantics of various type theories including MartinLof type theory, twolevel type theory and cubical type theory. We establish basic results in the semantics of type theory every type theory has a biinitial model; every model of a type theory has its internal language; the category of theories over a type theory is biequivalent to a full sub2category of the 2category of models of the type theory.

On Model Coding for Distributed Inference and Transmission in Mobile Edge Computing Systems ; Consider a mobile edge computing system in which users wish to obtain the result of a linear inference operation on locally measured input data. Unlike the offloaded input data, the model weight matrix is distributed across wireless Edge Nodes ENs. ENs have nondeterministic computing times, and they can transmit any shared computed output back to the users cooperatively. This letter investigates the potential advantages obtained by coding model information prior to ENs' storage. Through an informationtheoretic analysis, it is concluded that, while generally limiting cooperation opportunities, coding is instrumental in reducing the overall computationpluscommunication latency.

Randomly stirred perfect gas ; Foundations of the analysis of scaling in randomly stirred compressible fluid with the aid of stochastic differential equations are discussed in the example of perfect gas. The structure of the stress tensor with nonnegative shear and bulk viscosities is determined in ddimensional space. It is argued that the steady cascade picture of energy transfer is compatible with generic hydrodynamic equations. A renormalizable model of randomly stirred polytropic fluid is put forward and it is shown that this model should be used for description of randomly stirred perfect gas instead of the model of 'isothermal' fluid.

Global Wellposedness and Long Time Behaviors of ChemotaxisFluid System Modeling Coral Fertilization ; We consider generalized models on coral broadcast spawning phenomena involving diffusion, advection, chemotaxis, and reactions when egg and sperm densities are different. We prove the globalintime existence of the regular solutions of the models as well as their temporal decays in two and three dimensions. We also show that the total masses of egg and sperm density have positive lower bounds as time tends to infinity in three dimensions.

Boundaries and supercurrent multiplets in 3D LandauGinzburg models ; Theories with 3D mathcalN2 bulk supersymmetry may preserve a 2D mathcalN0,2 subalgebra when a boundary is introduced, possibly with localized degrees of freedom. We propose generalized supercurrent multiplets with bulk and boundary parts adapted to such setups. Using their structure, we comment on implications for the overlineQcohomology. As an example, we apply the developed framework to LandauGinzburg models. In these models, we study the role of boundary degrees of freedom and matrix factorizations. We verify our results using quantization.

Copula estimation for nonsynchronous financial data ; Copula is a powerful tool to model multivariate data. We propose the modelling of intraday financial returns of multiple assets through copula. The problem originates due to the asynchronous nature of intraday financial data. We propose a consistent estimator of the correlation coefficient in case of Elliptical copula and show that the plugin copula estimator is uniformly convergent. For nonelliptical copulas, we capture the dependence through Kendall's Tau. We demonstrate underestimation of the copula parameter and use a quadratic model to propose an improved estimator. In simulations, the proposed estimator reduces the bias significantly for a general class of copulas. We apply the proposed methods to real data of several stock prices.

Emergent Universality in a Quantum Tricritical Dicke Model ; We propose a generalized Dicke model which supports a quantum tricritical point. We map out the phase diagram and investigate the critical behaviors of the model through exact lowenergy effective Hamiltonian in the thermodynamic limit. As predicted by the Landau theory of phase transition, the order parameter shows nonuniversality at the tricritical point. Nevertheless, as a result of the separation of the classical and the quantum degrees of freedom, we find a universal relation between the excitation gap and the entanglement entropy for the entire critical line including the tricritical point. Here the universality is carried by the emergent quantum modes, whereas the order parameter is determined classically.

Reducing Anomaly Detection in Images to Detection in Noise ; Anomaly detectors address the difficult problem of detecting automatically exceptions in an arbitrary background image. Detection methods have been proposed by the thousands because each problem requires a different background model. By analyzing the existing approaches, we show that the problem can be reduced to detecting anomalies in residual images extracted from the target image in which noise and anomalies prevail. Hence, the general and impossible background modeling problem is replaced by simpler noise modeling, and allows the calculation of rigorous thresholds based on the a contrario detection theory. Our approach is therefore unsupervised and works on arbitrary images.

Ruin probabilities in the CramerLundberg model with temporarily negative capital ; We study the asymptotics of the ruin probability in the Cram'erLundberg model with a modified notion of ruin. The modification is as follows. If the portfolio becomes negative, the asset is not immediately declared ruined but may survive due to certain mechanisms. Under a rather general assumption on the mechanism satisfied by most such modified models from the literature we study the relation of the asymptotics of the modified ruin probability to the classical ruin probability. This is done under the Cram'er condition as well as for subexponential integrated claim sizes.

JUMT at WMT2019 News Translation Task A Hybrid approach to Machine Translation for Lithuanian to English ; In the current work, we present a description of the system submitted to WMT 2019 News Translation Shared task. The system was created to translate news text from Lithuanian to English. To accomplish the given task, our system used a Word Embedding based Neural Machine Translation model to post edit the outputs generated by a Statistical Machine Translation model. The current paper documents the architecture of our model, descriptions of the various modules and the results produced using the same. Our system garnered a BLEU score of 17.6.

Dengue model with earlylife stage of vectors and agestructure within host ; We construct an epidemic model for the transmission of dengue fever with an earlylife stage in the vector dynamics and agestructure within hosts. The earlylife stage of the vector is modeled via a general function that supports multiple vector densities. The it basic reproductive number and it vector demographic threshold are computed to study the local and global stability of the infectionfree state. A numerical framework is implemented and simulations are performed.

Model interactions for Pfaffian paired states based on ChernSimons field theory description ; On the basis of ChernSimons fieldtheoretical description we propose a simple method for derivation of model interactions for Pfaffian paired states. We verify the method in the case of Pfaffian i.e. Moore Read state, and derive a general form of the model interaction in the case of PH Pfaffian. More than one Landau level is needed to establish the correlations of the PH Pfaffian, and we present the values of relevant threebody pseudopotentials for two Landau levels.

A Preliminary Study on A Physical Model Oriented Learning Algorithm with Application to UAVs ; This paper provides a preliminary study for an efficient learning algorithm by reasoning the error from first principle physics to generate learning signals in near real time. Motivated by iterative learning control ILC, this learning algorithm is applied to the feedforward control loop of the unmanned aerial vehicles UAVs, enabling the learning from errors made by other UAVs with different dynamics or flying in different scenarios. This learning framework improves the data utilization efficiency and learning reliability via analytically incorporating the physical model mapping, and enhances the flexibility of the modelbased methodology with equipping it with the selflearning capability. Numerical studies are performed to validate the proposed learning algorithm.

Lattice approach to plane colorings ; I propose a fixedrange interaction multicomponent spin model, to be used as a physical analog to problems in plane geometry. Specifically, the model is applied to the open problem of the chromatic number of the plane. When spin values are interpreted as colors, the lowest energy configurations of the lattice spin system can be interpreted as approximations to plane colorings. In general minimum energy configurations of the model give optimal colorings, corresponding to minimum probability of any color realizing distance one. Approximate optimal lattice colorings with two to seven colors towards the continuum limit suggest that a true coloring of the plane cannot be achieved with less than seven colors.

On the geometry of magnetic Skyrmions on thin films ; We study the recently introduced 'critically coupled' model of magnetic Skyrmions, generalising it to thin films with curved geometry. The model feels keenly the extrinsic geometry of the film in threedimensional space. We find exact Skyrmion solutions on spherical, conical and cylindrical thin films. Axially symmetric solutions on cylindrical films are described by kinks tunnelling between 'vacua'. For the model defined on general compact thin films, we prove the existence of energy minimising multiSkyrmion solutions and construct the resolved moduli space of these solutions.

Does BERT agree Evaluating knowledge of structure dependence through agreement relations ; Learning representations that accurately model semantics is an important goal of natural language processing research. Many semantic phenomena depend on syntactic structure. Recent work examines the extent to which stateoftheart models for pretraining representations, such as BERT, capture such structuredependent phenomena, but is largely restricted to one phenomenon in English number agreement between subjects and verbs. We evaluate BERT's sensitivity to four types of structuredependent agreement relations in a new semiautomatically curated dataset across 26 languages. We show that both the singlelanguage and multilingual BERT models capture syntaxsensitive agreement patterns well in general, but we also highlight the specific linguistic contexts in which their performance degrades.

Investigating MetaLearning Algorithms for LowResource Natural Language Understanding Tasks ; Learning general representations of text is a fundamental problem for many natural language understanding NLU tasks. Previously, researchers have proposed to use language model pretraining and multitask learning to learn robust representations. However, these methods can achieve suboptimal performance in lowresource scenarios. Inspired by the recent success of optimizationbased metalearning algorithms, in this paper, we explore the modelagnostic metalearning algorithm MAML and its variants for lowresource NLU tasks. We validate our methods on the GLUE benchmark and show that our proposed models can outperform several strong baselines. We further empirically demonstrate that the learned representations can be adapted to new tasks efficiently and effectively.

Improving BackTranslation with Uncertaintybased Confidence Estimation ; While backtranslation is simple and effective in exploiting abundant monolingual corpora to improve lowresource neural machine translation NMT, the synthetic bilingual corpora generated by NMT models trained on limited authentic bilingual data are inevitably noisy. In this work, we propose to quantify the confidence of NMT model predictions based on model uncertainty. With word and sentencelevel confidence measures based on uncertainty, it is possible for backtranslation to better cope with noise in synthetic bilingual corpora. Experiments on ChineseEnglish and EnglishGerman translation tasks show that uncertaintybased confidence estimation significantly improves the performance of backtranslation.

An Empirical Study of Incorporating Pseudo Data into Grammatical Error Correction ; The incorporation of pseudo data in the training of grammatical error correction models has been one of the main factors in improving the performance of such models. However, consensus is lacking on experimental configurations, namely, choosing how the pseudo data should be generated or used. In this study, these choices are investigated through extensive experiments, and stateoftheart performance is achieved on the CoNLL2014 test set F0.565.0 and the official test set of the BEA2019 shared task F0.570.2 without making any modifications to the model architecture.

Estimating linear covariance models with numerical nonlinear algebra ; Numerical nonlinear algebra is applied to maximum likelihood estimation for Gaussian models defined by linear constraints on the covariance matrix. We examine the generic case as well as special models e.g. Toeplitz, sparse, trees that are of interest in statistics. We study the maximum likelihood degree and its dual analogue, and we introduce a new software package LinearCovarianceModels.jl for solving the score equations. All local maxima can thus be computed reliably. In addition we identify several scenarios for which the estimator is a rational function.

Quantumenhanced screened dark energy detection ; We propose an experiment based on a BoseEinstein condensate interferometer for strongly constraining fifthforce models. Additional scalar fields from modified gravity or higher dimensional theories may account for dark energy and the accelerating expansion of the Universe. These theories have led to proposed screening mechanisms to fit within the tight experimental bounds on fifthforce searches. We show that our proposed experiment would greatly improve the existing constraints on these screening models by many orders of magnitude, entirely eliminating the remaining parameter space of the simplest of these models.

Inclusion of the 3P0 model in PYTHIA 8 ; The spin effects in the hadronization process have been included for the first time in the PYTHIA 8 event generator. The spin effects are limited to the production of pseudoscalar mesons and are obtained from the propagation of the quark polarization along the fragmentation chain according to the rules of the socalled 3P0 model. The interface between PYTHIA 8 and the package of the 3P0 model is presented together with preliminary results on the Collins and dihadron asymmetries as obtained from simulations of the transversely polarized semiinclusive deep inelastic scattering process.

Antiadiabatic View of Fast Environmental Effects on Optical Spectra ; An antiadiabatic approach is proposed to model how the refractive index of the surrounding medium affects optical spectra of molecular systems in condensed phases. The approach solves some of the issues affecting current implementations of continuum solvation models and more generally of effective models where a classical description is adopted for the molecular environment.

Multiclass Multilingual Classification of Wikipedia Articles Using Extended Named Entity Tag Set ; Wikipedia is a great source of general world knowledge which can guide NLP models better understand their motivation to make predictions. Structuring Wikipedia is the initial step towards this goal which can facilitate finegrain classification of articles. In this work, we introduce the Shinra 5Language Categorization Dataset SHINRA5LDS, a large multilingual and multilabeled set of annotated Wikipedia articles in Japanese, English, French, German, and Farsi using Extended Named Entity ENE tag set. We evaluate the dataset using the best models provided for ENE label set classification and show that the currently available classification models struggle with large datasets using finegrained tag sets.

Global existence of a weak solution to unsaturated poroelasticity ; In this paper, we consider unsaturated poroelasticity, i.e., coupled hydromechanical processes in unsaturated porous media, modeled by a nonlinear extension of Biot's quasistatic consolidation model. The coupled, ellipticparabolic system of partial differential equations is a simplified version of the general model for multiphase flow in deformable porous media obtained under similar assumptions as usually considered for Richards' equation. In this work, the existence of a weak solution is established using regularization techniques, the Galerkin method, and compactness arguments. The final result holds under nondegeneracy conditions and natural continuity properties for the nonlinearities. The assumptions are demonstrated to be reasonable in view of geotechnical applications.

Dark matter and nonlocality of spacetime ; In this letter, we review a candidate for dark matter, known as OfDM, and explain how it relates to spacetime nonlocality. This connection provides a physical interpretation for why OfDM exists. Given the state of direct and indirect searches for dark matter, OfDM model would be an important candidate to consider, since it predicts all direct searches would fail to detect DM. The model has only one free parameter and in this regard, it is highly predictive. We review a few avenues to test this model.

The exact solution of a generalized twospins model ; We present the exact solution of a family of twospins models. The models are solved by the algebraic Bethe ansatz method using the gl2invariant Rmatrix and a multispins Lax operator. The interactions are by the Heisenberg spins exchange. We are also considering magnetic Bfields and a term for Haldane single spin anisotropy.

A Kinetically Constrained Model with a Thermodynamic Flavor ; Kinetically constrained models KCM generically have trivial thermodynamics and yet manifest rich glassy dynamics. In order to resolve the thermodynamicsdynamics disconnect in KCMs, we derive a KCM by coarsegraining a nontrivial thermodynamic model with a solvable partition function. Blending of thermodynamics to the KCM makes the role of landscape properties on the relaxation times, fragility and its crossover analytically transparent. Using Bethe ansatz, we calculate the timedependent Npoint probability function in the distinct dynamical regimes of the landscapes; the contribution of the glassy term is identified.

Ward Identities for the Standard Model Effective Field Theory ; We derive Ward identities for the Standard Model Effective Field Theory using the background field method. The resulting symmetry constraints on the Standard Model Effective Field Theory are basis independent, and constrain the perturbative and powercounting expansions. A geometric description of the field connections, and real representations for the rm SU2L times U1Y generators, underlies the derivation.

Sparse Representation of Gaussian Molecular Surface ; In this paper, we propose a model and algorithm for sparse representing Gaussian molecular surface. The original Gaussian molecular surface is approximated by a relatively small number of radial basis functions RBFs with rotational ellipsoid feature. The sparsity of the RBF representation is achieved by solving a nonlinear L1 optimization problem. Experimental results demonstrate that the original Gaussian molecular surface is able to be represented with good accuracy by much fewer RBFs using our L1 model and algorithm. The sparse representation of Gaussian molecular surface is useful in various applications, such as molecular structure alignment, calculating molecular areas and volumes, and the method in principle can be applied to sparse representation of general shapes and coarsegrained molecular modeling.

Cosmology in a model with Lagrange multiplier, and GaussBonnet and nonminimal kinetic couplings ; A scalartensor model with GaussBonnet and nonminimal kinetic couplings is considered, in which ghost modes are eliminated via a Lagrange multiplier constraint. A reconstruction procedure is deviced for the scalar potential and Lagrange multiplier, valid for any given cosmological scenario. In particular, inflationary and dark energy cosmologies of different types powerlaw, LittleRip, de Sitter, quasi de Sitter are reconstructed in such models. It is shown that, for various choices of the kinetic coupling terms, it is possible to obtain a viable inflationary phenomenology compatible with the most accurate values of the observational indices.

Modified Gravity in the framework of holographic dark energy ; The modified gravity is considered in the framework of the holographic dark energy. An analysis of the autonomous system, the critical points and their stability is presented. Unlike the dark energy models based on fR, it is found that working in the holographic frame enriches the possibility of accelerated and matter type points for different cosmological scenarios, making viable trajectories of successful fR models that are not allowed without the consideration of the holographic framework. The implications for the HuSawicki model are analyzed.

Data Ordering Patterns for Neural Machine Translation An Empirical Study ; Recent works show that ordering of the training data affects the model performance for Neural Machine Translation. Several approaches involving dynamic data ordering and data sharding based on curriculum learning have been analysed for the their performance gains and faster convergence. In this work we propose to empirically study several ordering approaches for the training data based on different metrics and evaluate their impact on the model performance. Results from our study show that prefixing the ordering of the training data based on perplexity scores from a pretrained model performs the best and outperforms the default approach of randomly shuffling the training data every epoch.

Timeconsistent decisions and rational expectation equilibrium existence in DSGE models ; Under some initial conditions, it is shown that time consistency requirements prevent rational expectation equilibrium REE existence for dynamic stochastic general equilibrium models induced by consumer heterogeneity, in contrast to static models. However, one can consider REEprohibiting initial conditions as limits of other initial conditions. The REE existence issue then is overcome by using a limit of economies. This shows that significant care must be taken of when dealing with rational expectation equilibria.

Tutorial on Implied Posterior Probability for SVMs ; Implied posterior probability of a given model say, Support Vector Machines SVM at a point bfx is an estimate of the class posterior probability pertaining to the class of functions of the model applied to a given dataset. It can be regarded as a score or estimate for the true posterior probability, which can then be calibratedmapped onto expected nonimplied by the model posterior probability implied by the underlying functions, which have generated the data. In this tutorial we discuss how to compute implied posterior probabilities of SVMs for the binary classification case as well as how to calibrate them via a standard method of isotonic regression.

A new neuralnetworkbased model for measuring the strength of a pseudorandom binary sequence ; Maximum order complexity is an important tool for measuring the nonlinearity of a pseudorandom sequence. There is a lack of tools for predicting the strength of a pseudorandom binary sequence in an effective and efficient manner. To this end, this paper proposes a neuralnetworkbased model for measuring the strength of a pseudorandom binary sequence. Using the Shrinking Generator SG keystream as pseudorandom binary sequences, then calculating the Unique Window Size UWS as a representation of Maximum order complexity, we demonstrate that the proposed model provides more accurate and efficient predictions measurements than a classical method for predicting the maximum order complexity.

Hyperfinite measurepreserving actions of countable groups and their model theory ; We give a shorter proof of a theorem of G. Elek stating that two hyperfinite measurepreserving actions of a countable group on standard probability spaces are approximately conjugate if and only if they have the same invariant random subgroup. We then use this theorem to study model theory of hyperfinite measurepreserving actions of countable groups on probability spaces. This work generalizes the modeltheoretic study of automorphisms of probability spaces conducted by I. Ben Yaacov, A. Berenstein, C. W. Henson and A. Usvyatsov.

Using Local Knowledge Graph Construction to Scale Seq2Seq Models to MultiDocument Inputs ; Querybased opendomain NLP tasks require information synthesis from long and diverse web results. Current approaches extractively select portions of web text as input to SequencetoSequence models using methods such as TFIDF ranking. We propose constructing a local graph structured knowledge base for each query, which compresses the web search information and reduces redundancy. We show that by linearizing the graph into a structured input sequence, models can encode the graph representations within a standard SequencetoSequence setting. For two generative tasks with very long text input, longform question answering and multidocument summarization, feeding graph representations as input can achieve better performance than using retrieved text portions.

New results on asymmetric thick branes ; This work deals with the presence and stability of thick brane solutions in the warped five dimensional braneworld scenario with a single extra spatial dimension of infinite extent. We combine two distinct procedures that give rise to new possibilities, allowing that we describe models of asymmetric thick branes, with the asymmetry being controlled by a single real parameter. We illustrate the main results with some distinct models, which show that the method works for both standard and generalized models, and the solutions are gravitationally stable against small perturbations of the metric.

Exponential Dynamical Localization for Random Word Models ; We show that onedimensional Schrodinger operators whose potentials arise by randomly concatenating words from an underlying set exhibit exponential dynamical localization EDL on any compact set which trivially intersects a finite set of critical energies. We do so by first giving a new proof of spectral localization for such operators and then showing that once one has the existence of a complete orthonormal basis of eigenfunctions with probability one, the same estimates used to prove it naturally lead to a proof of the aforementioned EDL result. The EDL statements provide new localization results for several classes of random Schrodinger operators including random polymer models and generalized Anderson models.

Modeling vehicular mobility patterns using recurrent neural networks ; Data on vehicular mobility patterns have proved useful in many contexts. Yet generative models which accurately reproduce these mobility patterns are scarce. Here, we explore if recurrent neural networks can cure this scarcity. By training networks on taxi from NYC and Shanghai, and personal cars from Michigan, we show most aspects of the mobility patterns can be reproduced. In particular, the spatial distributions of the street segments usage is well captured by the recurrent neural networks, which other models struggle to do.

Randommatrix perspective on manybody entanglement with a finite localization length ; We provide a simple and predictive randommatrix framework that naturally generalizes Page's law for ergodic manybody systems by incorporating a finite entanglement localization length. By comparing a highly structured onedimensional model to a completely unstructured model and a physical system, we uncover a remarkable degree of universality, suggesting that the effective localization length is a universal combination of model parameters up until it drops down to the microscopic scale.

Relative Entropy Method for the relaxation limit of Hydrodynamic models ; We show how to obtain general nonlinear aggregationdiffusion models, including KellerSegel type models with nonlinear diffusions, as relaxations from nonlocal compressible Eulertype hydrodynamic systems via the relative entropy method. We discuss the assumptions on the confinement and interaction potentials depending on the relative energy of the free energy functional allowing for this relaxation limit to hold. We deal with weak solutions for the nonlocal compressible Eulertype systems and strong solutions for the limiting aggregationdiffusion equations. Finally, we show the existence of weak solutions to the nonlocal compressible Eulertype systems satisfying the needed properties for completeness sake.

Estimating Centrality Blindly from Lowpass Filtered Graph Signals ; This paper considers blind methods for centrality estimation from graph signals. We model graph signals as the outcome of an unknown lowpass graph filter excited with influences governed by a sparse subgraph. This model is compatible with a number of data generation process on graphs, including stock data and opinion dynamics. Based on the said graph signal model, we first prove that the folklore heuristics based on PCA of data covariance matrix may fail when the graph filter is not sufficiently lowpass. To remedy, we propose a robust blind centrality estimation method which substantially improves the centrality estimation performance. Numerical results on synthetic and real data support our findings.

Rate Distortion Study for TimeVarying Autoregressive Gaussian Process ; Ratedistortion formulation is the informationtheoretic approach to the study of signal encoding systems. Since a more general approach to model the nonstationarity exhibited by realworld signals is to use appropriately fitted time varying autoregressive TVAR models, we have investigated the ratedistortion function for the class of time varying nonstationary signals. In this study, we present formulations of the ratedistortion function for the Gaussian TVAR processes. The ratedistortion function can serve as an informationtheoretic bound on the performance achievable by source encoding techniques when the processing signal is represented exclusively by a Gaussian TVAR model.

Transfer Learning from Transformers to Fake News Challenge Stance Detection FNC1 Task ; In this paper, we report improved results of the Fake News Challenge Stage 1 FNC1 stance detection task. This gain in performance is due to the generalization power of large language models based on Transformer architecture, invented, trained and publicly released over the last two years. Specifically 1 we improved the FNC1 best performing model adding BERT sentence embedding of input sequences as a model feature, 2 we finetuned BERT, XLNet, and RoBERTa transformers on FNC1 extended dataset and obtained stateoftheart results on FNC1 task.

Selective Bootstrap Percolation ; A new class of bootstrap percolation models in which particle culling occurs only for certain numbers of nearest neighbours is introduced and studied on a Bethe lattice. Upon increasing the density of initial configuration they undergo multiple hybrid or mixedorder phase transitions, showing that such intriguing phase behaviours may also appear in fully homogeneous situationsenvironments, provided that culling is selective rather than cumulative. The idea immediately extends to facilitation dynamics, suggesting a simple way to construct onecomponent models of multiple glasses and glassglass transitions as well as more general coarsegrained models of complex cooperative dynamics.

Majorana dark matter and neutrino mass with S3 symmetry ; This model includes a minimal extension of the standard model with S3 and Z2 symmetries to explain neutrino masses and mixing along with the dark matter phenomenology. Neutrino phenomenology is explored, consistent with the 3 sigma observation of oscillation parameters and a nonzero reactor mixing angle theta13 is obtained. The S3 singlet Majorana neutrino couples to the third generation of leptons, gives a correct relic density compatible with the Planck data. This model does not allow tree level direct detection, therefore we discuss the loop level effective interaction with the nucleus mediated by gauge boson. Also the constraints from the lepton flavor violating rare decay mode is commented.

KaluzaKlein FRW type dark energy model with a massive scalar field ; In this investigation we discuss the dynamical aspects of KaluzaKlein FRW type cosmological model in the presence of dark energy fluid and an attractive massive scalar field. The field equations are solved using a power law between the average scale factor and the scalar field to reduce the mathematical complexity. We have presented the corresponding dark energy model. Important cosmological parameters like equation of state EoS parameter, the deceleration parameter, the density parameter and the jerk parameter are evaluated. The physical behavior of the parameters is also discussed.

Surrogate Modeling of Dynamics From Sparse Data Using Maximum Entropy Basis Functions ; In this paper we present a data driven approach for approximating dynamical systems. A dynamics is approximated using basis functions, which are derived from maximization of the informationtheoretic entropy, and can be generated directly from the data provided. This approach has advantages over other methods, where a dictionary of basis functions have to be provided by the user, which is non trivial in some applications. We compare the accuracy of the proposed datadriven modeling approach to existing methods in the literature, and demonstrate that for some applications the maximum entropy basis functions provide significantly more accurate models.

Characterbased NMT with Transformer ; Characterbased translation has several appealing advantages, but its performance is in general worse than a carefully tuned BPE baseline. In this paper we study the impact of characterbased input and output with the Transformer architecture. In particular, our experiments on ENDE show that characterbased Transformer models are more robust than their BPE counterpart, both when translating noisy text, and when translating text from a different domain. To obtain comparable BLEU scores in clean, indomain data and close the gap with BPEbased models we use known techniques to train deeper Transformer models.

Numerical solution of a bendingtorsion model for elastic rods ; Aiming at simulating elastic rods, we discretize a rod model based on a general theory of hyperelasticity for inextensible and unshearable rods. After reviewing this model and discussing topological effects of periodic rods, we prove convergence of the discretized functionals and stability of a corresponding discrete flow. Our experiments numerically confirm thresholds e.g. for Michell's instability and indicate a complex energy landscape, in particular in the presence of impermeability.

Unitary toy qubit transport model for black hole evaporation ; In a recent paper Osuga and Page have presented an explicitly unitary toy qubit transport model for transferring information from a black hole to the outgoing radiation. Following their idea we propose a unitary toy model which involves fermionic Hawking states.

A Random Clockwork of Flavor ; We propose a simple clockwork model of flavor which successfully generates the Standard Model flavor hierarchies from random orderone couplings. With very few parameters we achieve distributions of models in excellent agreement with observation. We explain in some detail the interpretation of our mechanism as random localization of zero modes in theory space. The scale of the vectorlike fermions is mostly constrained by lepton flavor violation with secondary constraints arising from rare meson decays.

Improving Conditioning in ContextAware Sequence to Sequence Models ; Neural sequence to sequence models are well established for applications which can be cast as mapping a single input sequence into a single output sequence. In this work, we focus on cases where generation is conditioned on both a short query and a long context, such as abstractive question answering or documentlevel translation. We modify the standard sequencetosequence approach to make better use of both the query and the context by expanding the conditioning mechanism to intertwine query and context attention. We also introduce a simple and efficient data augmentation method for the proposed model. Experiments on three different tasks show that both changes lead to consistent improvements.

Instability of the big bang coordinate singularity in a Milnelike universe ; We present a simplified dynamicvacuumenergy model for a timesymmetric Milnelike universe. The big bang singularity in this simplified model, like the one in a previous model, is just a coordinate singularity with finite curvature and energy density. We then calculate the dynamic behavior of scalar metric perturbations and find that these perturbations destabilize the big bang singularity.

Scalarfield potential for viable models in fR theory ; The fR theory of gravity can be expressed as a scalar tensor theory with a scalar degree of freedom phi. By a conformal transformation, the action and its GibbonsYorkHawking boundary term are written in the Einstein frame and the field equations are found. An effective potential is defined from part of the trace of the field equations in such a way that it can be calculated as an integral of a purely geometric term. This potential as well as the scalar potential are found, plotted and analyzed for some viable models of fR and for two other proposed new, shown viable, models.

TwoStage Learning for Uplink Channel Estimation in OneBit Massive MIMO ; We develop a twostage deep learning pipeline architecture to estimate the uplink massive MIMO channel with onebit ADCs. This deep learning pipeline is composed of two separate generative deep learning models. The first one is a supervised learning model and designed to compensate for the quantization loss. The second one is an unsupervised learning model and optimized for denoising. Our results show that the proposed deep learningbased channel estimator can significantly outperform other stateoftheart channel estimators for onebit quantized massive MIMO systems. In particular, our design provides 510 dB gain in channel estimation error. Furthermore, it requires a reasonable amount of pilots, on the order of 20 per coherence time interval.

Limits on 331 vector bosons from LHC proton collision data ; In this paper, limits are set on bileptons masses and couplings in the context of the the 331 Model, as well as generic, non331 Models predicting bileptons. The following measurable processes are studied pp rightarrow ellellellellX, pp rightarrow ellell nu nu X and pp rightarrow ellell X. Experimental limits on singlycharged bileptons masses and couplings within 331 Models are also obtained for the first time. With the results, an over 20 yearold experimental limit on vector bileptons is increased by 60. The computed limits are now the most stringent ones for these particles.

Object Detection with Convolutional Neural Networks ; In this chapter, we present a brief overview of the recent development in object detection using convolutional neural networks CNN. Several classical CNNbased detectors are presented. Some developments are based on the detector architectures, while others are focused on solving certain problems, like model degradation and smallscale object detection. The chapter also presents some performance comparison results of different models on several benchmark datasets. Through the discussion of these models, we hope to give readers a general idea about the developments of CNNbased object detection.

Compactified Jacobians as Mumford models ; We show that relative compactified Jacobians of oneparameter smoothings of a nodal curve of genus g are Mumford models of the generic fiber. Each such model is given by an admissible polytopal decomposition of the skeleton of the Jacobian. We describe the decompositions corresponding to compactified Jacobians explicitly in terms of the auxiliary stability data and find, in particular, that in degree g there is a unique compactified Jacobian encoding slop stability, and it is induced by the tropical break divisor decomposition.

Hidden Sector and Gravity ; In this paper, we consider a generic hidden sector which interacts only gravitationally with Standard Model particles. We show that quantum gravity leads to operators which can be probed with fifth force type experiments. The EotWash torsion pendulum experiment implies that the masses of any scalar field or any massive spin2 field that couples with the usual gravitational strength to the energymomentum tensor of the Standard Model must be larger than 103eV. This has interesting consequences for models of dark matter which posit very light scalar fields. Dark matter must be heavier than 103eV if it is a scalar field or a massive spin2 field.

Tumor ablation due to inhomogeneous anisotropic diffusion in generic 3dimensional topologies ; We derive a full 3dimensional 3D model of inhomogeneous anisotropic diffusion in a tumor region coupled to a binary population model. The diffusion tensors are acquired using Diffusion Tensor Magnetic Resonance Imaging DTI from a patient diagnosed with glioblastoma multiform GBM. Then we numerically simulate the full model with Finite Element Method FEM and produce drug concentration heat maps, apoptosis regions, and doseresponse curves. Finally, predictions are made about optimal injection locations and volumes, which are presented in a form that can be employed by doctors and oncologists.

Choi states, symmetrybased quantum gate teleportation, and storedprogram quantum computing ; The storedprogram architecture is canonical in classical computing, while its power has not been fully recognized for the quantum case. We study quantum information processing with stored quantum program states, i.e., using qubits instead of bits to encode quantum operations. We develop a storedprogram model based on Choi states, following from channelstate duality, and a symmetrybased generalization of deterministic gate teleportation. Our model enriches the family of universal models for quantum computing, and can also be employed for tasks including quantum simulation and communication.

Groundstate energy of a RichardsonGaudin integrable BCS model ; We investigate the groundstate energy of a RichardsonGaudin integrable BCS model, generalizing the closed and open pip models. The Hamiltonian supports a family of mutually commuting conserved operators satisfying quadratic relations. From the eigenvalues of the conserved operators we derive, in the continuum limit, an integral equation for which a solution corresponding to the ground state is established. The energy expression from this solution agrees with the BCS meanfield result.

Sigmoidal Inflation ; In this paper we present a new cosmological inflationary model which is constructed using the IvanovSalopekBond method with a logistic generating function. We derive the inflationary observables as well as the duration and temperature of the subsequent reheating epoch of our model exactly, with no need to recur to the slow roll approximation. The obtained scalar spectral index and tensortoscalar ratio of perturbations fall comfortably within the range of the measurements presented by the Planck collaboration. On the other hand, for the reheating era, our model predicts a relatively small number of efolds and thus high temperatures, still within range of Planck's bounds.

SGVAE Sequential Graph Variational Autoencoder ; Generative models of graphs are wellknown, but many existing models are limited in scalability and expressivity. We present a novel sequential graphical variational autoencoder operating directly on graphical representations of data. In our model, the encoding and decoding of a graph as is framed as a sequential deconstruction and construction process, respectively, enabling the the learning of a latent space. Experiments on a cycle dataset show promise, but highlight the need for a relaxation of the distribution over node permutations.

Effective Integration of Symbolic and Connectionist Approaches through a Hybrid Representation ; In this paper, we present our position for a neuralsymbolic integration strategy, arguing in favor of a hybrid representation to promote an effective integration. Such description differs from others fundamentally, since its entities aim at representing AI models in general, allowing to describe both nonsymbolic and symbolic knowledge, the integration between them and their corresponding processors. Moreover, the entities also support representing workflows, leveraging traceability to keep track of every change applied to models and their related entities e.g., data or concepts throughout the lifecycle of the models.

A reduced model for solute transport in compliant blood vessels with arbitrary axial velocity profile ; We derive a reduced model of solute transport in blood based on the center manifold theory. The derivation is carried out on a convection diffusion equation with general axial and radial velocity profiles in a blood vessel of varying cross section. We couple the resulting one dimensional equation to a reduced model for blood flow in a compliant vessel. In the special case of a noslip axial velocity profile, we study the dependence of the diffusion coefficient and corresponding numerical solutions on the shape of the profile.

Towards Symbolic Factual Change in DEL ; We extend symbolic model checking for Dynamic Epistemic Logic DEL with factual change. Our transformers provide a compact representation of action models with pre and postconditions, for both S5 and the general case. The method can be implemented using binary decision diagrams and we expect it to improve model checking performance. As an example we give a symbolic representation of the SallyAnne false belief task.

Inflation with very small tensortoscalar ratio ; We have investigated inflation models that predict a very small value of the tensortoscalar ratio, r. The spectral index ns, and the tensortoscalar ratio r, are strictly constrained by the Planck data. ns and r are sensitive to the shape and magnitude of the inflaton potential, respectively.The constraints by the Planck 2018 data combined with other cosmological observations are compared with the predictions from the inflation models regarding ns and r. Furthermore, we discuss the comparison of future tensortoscalar ratio data with predictions from the inflation models with a focus on part of the quantum fluctuation origin.

Multiview Representation Learning for a Union of Subspaces ; Canonical correlation analysis CCA is a popular technique for learning representations that are maximally correlated across multiple views in data. In this paper, we extend the CCA based framework for learning a multiview mixture model. We show that the proposed model and a set of simple heuristics yield improvements over standard CCA, as measured in terms of performance on downstream tasks. Our experimental results show that our correlationbased objective meaningfully generalizes the CCA objective to a mixture of CCA models.

Modeling Neural Architecture Search Methods for Deep Networks ; There are many research works on the designing of architectures for the deep neural networks DNN, which are named neural architecture search NAS methods. Although there are many automatic and manual techniques for NAS problems, there is no unifying model in which these NAS methods can be explored and compared. In this paper, we propose a general abstraction model for NAS methods. By using the proposed framework, it is possible to compare different design approaches for categorizing and identifying critical areas of interest in designing DNN architectures. Also, under this framework, different methods in the NAS area are summarized; hence a better view of their advantages and disadvantages is possible.

Causal Mosaic CauseEffect Inference via Nonlinear ICA and Ensemble Method ; We address the problem of distinguishing cause from effect in bivariate setting. Based on recent developments in nonlinear independent component analysis ICA, we train nonparametrically general nonlinear causal models that allow nonadditive noise. Further, we build an ensemble framework, namely Causal Mosaic, which models a causal pair by a mixture of nonlinear models. We compare this method with other recent methods on artificial and real world benchmark datasets, and our method shows stateoftheart performance.

Extended analysis for the Evolution of the Cosmological history in Einsteinaether Scalar Field theory ; We consider an Einsteinaether scalar field cosmological model where the aether and the scalar field are interacting. The model of our consideration consists the two different interacting models proposed in the literature by Kanno et al. and by Donnelly et al. We perform an extended analysis for the cosmological evolution as it is provided by the field equations by using methods from dynamical systems; specifically, we determine the stationary points and we perform the stability analysis of those exact solutions.

Covariant quantum corrections to a scalar field model inspired by nonminimal natural inflation ; We calculate the covariant oneloop quantum gravitational effective action for a scalar field model inspired by the recently proposed nonminimal natural inflation model. Our calculation is perturbative, in the sense that the effective action is evaluated in orders of background field, around a Minkowski background. The effective potential has been evaluated taking into account the finite corrections. An orderofmagnitude estimate of the oneloop corrections reveals that gravitational and nongravitational corrections have same or comparable magnitudes.

Closed Form Solutions For The Quantum LRS Bianchi IX and VIII Wheeler DeWitt Equations For An Arbitrary HartleHawking Ordering Parameter ; In this paper we present the quantum LRS Bianchi IX models or, as they may be better known, the TaubNUT models in addition to the quantum LRS Bianchi VIII models, both coupled to a stiff matter source. We solve their Wheeler DeWitt equations for an arbitrary HartleHawking ordering parameter using separation of variables. Afterwards we construct a superposition of their respective wave functions and discuss some of the interesting qualitative properties they possess.